Explorations in Interactive Visual Analytics
summary
Built an interactive platform called IV Workbench — a distributed vector functional column store that pulls a 100 TB NYSE-style feed down to a workable 35 million records and lets analysts iterate in minutes. Rendering moves to the server: server-side bitmap rendering cuts a million-point plot from ~22s (browser) to ~4s; 500M plots run ~116s but drop to ~27s with small scripting optimizations. Using binning, opacity and pre-attentive cues the workflow finds needles (≈2,000 points in 1M) and a PageRank hit on node 200 as the suspect.
AI-written summary · about
chapters (39)
- 0:00Introduction
- 0:26Agenda
- 1:05Big Data
- 2:14Data Scientist
- 3:48IV Workbench
- 5:44Challenges
- 7:10Case Study
- 16:08Exploring Relationships
- 16:50What We Learned
- 17:20Integrated Analysis Visualization Environment
- 17:45Data Visualization
- 18:40The Latch Model
- 19:55Leverage Serverside Processing
- 20:05Example Case Study
- 21:31Rendering on Clients
- 23:44Pattern Matching
- 23:57Analysis Techniques
- 26:03Aggregation
- 26:40Opacity
- 26:57Scatter
- 27:07Heatmap
- 27:17Histogram
- 28:09Results
- 28:55Visual Design
- 29:09Edward Tufte
- 31:47The 7 Rule
- 32:17Top 80 Keywords
- 32:36Gaps
- 33:06Filters
- 33:25Plot Ratio
- 34:19Hiding Controls
- 35:12Data Complexity
- 35:42Large Data Sets
- 36:29Query Tool
- 37:41Auto Updates
- 38:27Execution
- 39:20Scripting
- 41:30Analyst Interface
- 45:17Summary
related talks
42mInteractive Data Analysis: Visualization and Beyond
30mSeeing the needle AND the haystack: single-datapoint selection for billion-poin...
36mMaking interactive visualizations easy, inside and outside of notebooks
41mVisualizing big data
27mScalable Data Visualization
43mAdvanced Data Visualizations In Jupyter Notebooks
30mData @Scale Seattle- Pawel Terlecki
28mAdvanced Graph Visualizations in Jupyter Notebooks